Describe the bug
First noticed in #1968, when using any of the built-in recipes which use aggregate.ensure_aggregatable_across_cases on operational data pulled from MASS (.pp format), it fails with error
"Cube should have 'forecast_period' and 'forecast_reference_time' dimension coordinates."
This is as a result of read.read_cubes not loading these dimension coordinates when loading in the cubes.
How to reproduce
Either by workflow or command line running cset bake, choose an aggregation method which uses ensure_aggregatable_across_cases on archived MASS data.
Expected behaviour
This isn't a case of missing metadata, the forecast_period, forecast_reference_time and time coordinates all exist in the data, but do not exist as dimension coordinates.
The RES is able to aggregate with this data, so it should be possible in CSET too.
Describe the bug
First noticed in #1968, when using any of the built-in recipes which use
aggregate.ensure_aggregatable_across_caseson operational data pulled from MASS (.pp format), it fails with errorThis is as a result of
read.read_cubesnot loading these dimension coordinates when loading in the cubes.How to reproduce
Either by workflow or command line running cset bake, choose an aggregation method which uses
ensure_aggregatable_across_caseson archived MASS data.Expected behaviour
This isn't a case of missing metadata, the
forecast_period,forecast_reference_timeandtimecoordinates all exist in the data, but do not exist as dimension coordinates.The RES is able to aggregate with this data, so it should be possible in CSET too.